The phenomenon of online diffusion of misattributed quotes is so widespread that got its own dedicated meme. You may have seen a picture of Abraham Lincoln, the 16th President of the United States, that warns: ‘Don’t believe everything you read on the Internet just because there’s a picture with a quote next to it’. Lincoln, apparently nicknamed ‘Honest Abe’ when young, was assassinated in 1865, which makes it unlikely he had opinions about the Internet, and he is one of the historical celebrities most quoted (often incorrectly) on the web. Lincoln shares this questionable honour with the likes of Mark Twain and Albert Einstein. ‘The definition of insanity is doing the same thing over and over again and expecting a different result’ is one of the most famous quotes of Albert Einstein. Except that is not: the earliest known exact match of the quote appears in a Narcotics Anonymous information pamphlet in 1981, some 25 years after Einstein’s death.
Intuitively, the association with a prestigious author should amplify the credibility and appeal of a quote, and this may explain the success of misattributed quotes on the web. Or, at least, this is what Jamie Tehrani and myself thought. To test the hypothesis we selected 60 mildly famous, but not, according to us, recognizable quotes in three different domains: love/friendship (such as “If you love somebody, set them free”), money/success (e.g. “There are people who have money and people who are rich”), and science/literature (e.g. “Science may set limits to knowledge, but should not set limits to imagination”). Since we believed that the association with famous authors would have had an effect, we wanted to understand if the domain modulated this effect. We reasoned that the influence would have been stronger if the author was famous because of the activity in the specific domain, thus famous scientists and writers (‘Charles Darwin’, ‘Edgar Allan Poe’) would have been particularly preferred in the science/literature domain. We expected less influence for the domain money/success—famous authors may have some general experience of personal success, more than common people, but that’s not, so to say, their specialty—and even less for the domain love/friendship, where one would not expect famous authors to know necessarily more than any of us.
The participants of the experiment were supposed to help us to choose the ‘most inspirational quotes’ through a web interface, and they were presented with pairs of quotes from the same domain, one associated with a famous author, and one with a randomly generated name (‘Odis Pennel’, ‘Romeo Lyon’, and similar). We asked them to choose the one they preferred (notice that, even if we tried to use wide-ranging quotes, this randomization may have created in few cases some suspects in the participants: few lucky ones may have seen for example the quote ‘A kiss is a lovely trick, designed by nature, to stop words when speech becomes unnecessary’ credited to Max Weber or Mahatma Gandhi…)
Long story short: we did not find any effect of the association with famous authors. It did not seem to make any difference if a quote was associated with Albert Einstein or to Romeo Lyon, no matter what the domain was. Of course, the first thing we thought was that we had made some errors or that participants, that were recruited online, were simply choosing randomly one quote or the other. What we found, however, was that the success of a quote was correlated with the success of the same quote in a control condition, were pairs of aphorisms were presented without any indication of their (supposed) source. In sum, participants were actually reading the quotes and evaluating their content, and not who (supposedly) said them. Quotes like ‘It is better to have loved and lost, then never to have loved at all’ had consistently more success than, for example, ‘love is the big booming beat which covers up the noise of hate’, independently from whom was associated with the quote.
Similarly, we did not find any conformity effect. We run a condition of the experiment in which, instead of ‘famous’ and ‘unknown’ authors, we associated to the quotes a previous number of ‘likes’ (or the number of participants that had already chosen them). We ensured that the number of likes of the ‘popular’ quote was around three times more than the number of likes of the ‘unpopular’ one, so, for example, one quote was liked around 300 times, and the other one around 100 (or the 75% and the 25% of the total number of likes – numbers were randomised each time). According to the cultural evolution definition, the probability of conformist individuals to prefer the popular quote should be higher than the actual proportion, so higher than 75%. We found that participants were more likely to choose the popular quote, but, in fact, less than 75% of times. There was, thus, an effect of popularity (again modulated by the actual content of the quote), but not an effect of conformity. In the figure below, the proportions of time the popular and unpopular quotes had been preferred should have been in the grey area in case of conformity (and should have been both around 50% in case of no effect of popularity.
Contrary to our expectations, the association with prestigious authors did not boost the preferences for the quotes. Why, then, the diffusion of misattributed quotes? We speculated that the reason could be, in fact, the opposite of the intuitive one. A good quote will be more likely to succeed independently from whether it is associated with a famous author or not. However, when reproduced again and again, the information on the author—which is seemingly less important than the content itself—could get lost. This may favour famous authors in two ways: easy-to-remember authors will replace authors that are more difficult to remember, such as, for example, the quote ‘Your assumptions are your windows on the world. Scrub them off every once in a while, or the light won’t come in’, which is usually credited to Isaac Asimov, while the words were pronounced by Alan Alda, a less famous American actor and director. Second, anonymous quotes will be more likely to be credited to well-known individuals than to lesser known, like ‘Be Yourself, Everyone else is already taken’, an aphorism for which we have only very recent matches, but misattributed to Oscar Wilde.
In sum, while the final result is the same—famous authors and successful quotes go hand in hand—the causal relationship is reversed: the success of quotations would not be the result of being misattributed to famous authors. On the contrary, misattributions would be the result of the wide diffusion of good quotes.
Also, some more general considerations for cultural evolution, from the discussion section:
Our results contribute to a growing body of works that found contrasting result on the effects of context-based biases […] Establishing the relative importance of context and content biases, for cultural evolutionary studies, is a task that goes beyond the mere need for terminological precision. Context-based biases are relatively simple, domain-general, heuristics. If they are the main driving force of cultural evolution, cultural evolution studies should mainly focus on population-level dynamics. Modelling strategies, or theoretical approaches, in which the cognitive properties of human individuals are only minimally sketched will do the job. On the contrary, content-based biases depend on domain-specific cognitive aspects, and, if the success of practices and ideas depend mostly on those, cultural evolutionists need to pay particular attention to the subtleties of human cognition.